FuseFX breaks down the water sims and comp work involved.
M. Night Shyamalan’s Knock at the Cabin includes a spectacular tsunami sequence witnessed on television by a family held hostage in a remote rural locale.
The visual effects for the apocalyptic scenes, which occur near Haystack Rock in Oregon, were crafted by FuseFX. The VFX studio needed to deal with hand-held cameras and complex water sims in record time, as visual effects supervisor Tommy Tran tells befores & afters.
b&a: Where did FuseFX get started with this tsunami sequence?
Tommy Tran: When we came on board, that was in previs land and they said, ‘We have four to six weeks to get it done. Can you do it?’. We said, ‘Yes, we can.’ We got a small team together knowing that it was going to be a big challenge. At the time, it was only two shots, just the tsunami as an establishing shot, and then the full shot where we created the big wave. It was a small team, but they were all heavy hitters. Everybody that was assigned was a senior, if not supervisor level. Lean and mean is how we went on this.
b&a: What were the first steps?
Tommy Tran: We looked up the location where the live-action had been shot, we mapped it out geographically and we took measurements of the real rocks. So then we started developing the shots and we showed them our first couple of rounds of tests.
There were a lot of extras on the day, but some of them weren’t really paying attention. So we had to remove a lot of those people and replace them with digital doubles. That was fun. You would never know that we took over a lot of those people in the background. They were literally two or three pixels tall that got enveloped in the wave. We needed the interaction of them falling and tumbling, whereas the real actors just ran to the left, they never got hit by the wave.
We spent a lot of time and resources with Golaem, our crowd simulator, to replace the two pixel actors in the deep background so that they would interact and they would get engulfed and actually get sucked up as the wave came through.
b&a: Oh wow, that’s interesting. What about people closer to camera?
Tommy Tran: There was a family at the end of the crash of the wave that we digi-doubled, too. Led by our CG Supervisor, Shirak Agresta, we developed and then match-moved them so that when the wave actually came, it wrapped around their feet and pushed them over, they fell, they tumbled, and then you see their bodies as ghostly apparitions in the wall of the wave.
b&a: So how did Fuse FX approach the wave itself?
Tommy Tran: The first shot was the establishing shot, and it was just the wave starting to build on the horizon. That was the simpler of the two setups. We would sim the wave and we let it run long, and then we used the front part, before it crested and crashed, for the establishing shot.
Once the crash happened, we had to take that to another level because of all the layers that were involved, like the water receding back as the crest happens. In reality, the wave happens when the full momentum of the wave touches the ground. It sucks all the water away from the sand. So we had to do simulations for that, the receding water, we had to do the foam, the sand that’s picked up, which changes the color in the foam, and then the interaction with the people.
Once we had the digi-doubles match-moved to the plate, then we had to do simulations of water splashing off of them so that they didn’t just feel like they were falling and just dissolved into the wave.
Then there’s the rock smash. That was beautiful and epic. That made the shot. If it wasn’t for that rock, it would’ve just been another wave crashing. Since we had those real rocks in CG space, the way that the FX team wrapped the water around it and sprayed all the spray around it–we even broke off little bits of rocks–made that shot to me one of the more epic shots in the last couple of years. The sim on that took weeks just to get it right.
In fact, there were so many elements that made up that final wave that we ended up rendering it in deep. We used deep in order to hold out certain elements from one another without having a baked-in matte. It gives you so much control over holding out things in z-depth, but it is heavy. It adds an exponential amount of time to the renders and comps, but sometimes you have to bite the bullet and go that route so that you don’t re-render every time you want to move a digi-double to the left or put them deeper into the water.
Most of the credit has to go to the effects team, Samantha Williams and Tyler Britton. And to my VFX producer, Delane Leahy. She was instrumental in the communication with the client, including Jeff Robinson, their visual effects producer. It was a real pleasure working with them, including M. Knight.
b&a: It’s also handheld for most of the shot, that must have been a challenge just in and of itself?
Tommy Tran: Oh, yeah. The tracking, we had to re-track it several times just to make sure that the horizon was right. We didn’t have all the lens information, which made it a little more difficult. But it’s the year 2023. There’s nothing that anyone can’t track, even without proper camera information.
b&a: When the people are shown tumbling in the wave, the camera’s also tumbling. How did you get the right look there?
Tommy Tran: That was left to us. The discussion in our meetings was, ‘Okay, if you’re standing with a GoPro where the waves are breaking, what would happen to your camera?’ You always want to lock on the actor. You never want to lose the actor’s face, but if you’re getting hit by a wave, you’re going to tumble. So we’re like, ‘How many tumbles does the camera do? One? Two? And then how much motion blur is associated with each tumble?’
So we went from three, to one, to one and a half, until we settled on the tumble that gave the audience the perception that we were out of control when we got into the wave, but then we had to keep the rotation small enough so that we didn’t lose everything to motion blur, so that you can actually see some bubbles in the most chaotic moments of getting hit.
